Psychophysiology, 46 (2009), 957–961. Wiley Periodicals, Inc. Printed in the USA.
Copyright r 2009 Society for Psychophysiological Research
DOI: 10.1111/j.1469-8986.2009.00848.x
The stability of error-related brain activity with
increasing trials
DOREEN M. OLVET AND GREG HAJCAK
Department of Psychology, Stony Brook University, Stony Brook, New York, USA
Abstract
The error-related negativity (ERN) and error positivity (Pe) are increasingly being examined as neural correlates of
response monitoring. The minimum number of error trials included in grand averages varies across studies; indeed,
there has not been a systematic investigation on the number of trials required to obtain a stable ERN and Pe. In the
current study, the ERN and Pe were quantified as two random trials were added to participants’ (N 5 53) ERP
averages. Adding trials increased the correlation with the grand average ERN and Pe; however, high correlations
(rs4.80) were obtained with only 6 trials. Internal reliability of the ERN and Pe reached moderate levels after 6 and 2
trials and the signal-to-noise ratio of the ERN and Pe did not change after 8 and 4 trials, respectively. Combined, these
data suggest that the ERN and Pe can be quantified using a minimum of between 6 and 8 error trials.
Descriptors: Cognition, EEG/ERP, Normal volunteers
The ERN and Pe have been investigated in a variety of withinsubject (Hajcak & Foti, 2008; Hajcak, Moser, Yeung, & Simons,
2005; Nieuwenhuis et al., 2001; Pailing & Segalowitz, 2004;
Ullsperger & von Cramon, 2006) and between-subject designs
(Band & Kok, 2000; Franken, van Strien, Franzek, & van de
Wetering, 2007; Hajcak, Franklin, Foa, & Simons, 2008; Luu,
Collins, & Tucker, 2000; Mathalon et al., 2002). In these experiments, participants typically perform between 250 and 1,500
trials of a speeded reaction time task in relatively rapid succession. Errors tend to be rare, however, resulting in a relatively low
number of trials in some ERP averages. In fact, the minimum
number of error trials for conditions and participants
varies greatly, ranging from 5 to 300 (e.g., Amodio et al., 2004;
Franken et al., 2007; Hajcak & Simons, 2008; Morris, Yee, &
Nuechterlein, 2006; Ullsperger & von Cramon, 2006; Zirnheld
et al., 2004). It has been suggested that only a few error trials
are required for a stable ERN (see Hajcak & Simons, 2008);
however, guidance on the actual number of trials required
to obtain a stable ERN based on empirical work is lacking
(for similar work on the P130, cf. Cohen & Polich, 1997; Polich,
1986).
In the present study, we set forth to systematically assess the
stability of the ERN and the Pe as an increasing number of errors
trials were examined. We used methods similar to that reported
by Polich and colleagues (Cohen & Polich, 1997; Polich, 1986).
First, we measured the mean and standard deviation of the ERN
and Pe as random pairs of error trials were added to participant
averages (i.e., 2, 4, 6, 8, 10, 12, and 14 error trials). Additionally,
we calculated the correlation between these averages and the
ERN/Pe for all error trials (i.e., grand average). Internal reliability of the ERN and Pe as a function of increasing trial
numbers was quantified with Cronbach’s alpha; finally, the
The error-related negativity (ERN) is an event-related potential
(ERP) that presents as a negative deflection approximately 50 ms
following an erroneous response at fronto-central midline recording sites (Falkenstein, Hohnsbein, Hoormann, & Blanke,
1991; Gehring, Goss, Coles, Meyer, & Donchin, 1993). Functionally, the ERN appears to reflect relatively early error processing in the medial prefrontal cortex (cf. Ridderinkhof,
Ullsperger, Crone, & Nieuwenhuis, 2004), and source localization studies suggest that the ERN is generated in the anterior
cingulate cortex (ACC; Dehaene, Posner, & Tucker, 1994; Holroyd, Dien, & Coles, 1998; van Veen & Carter, 2002). Computational models propose that the ERN represents a
reinforcement learning signal (Holroyd & Coles, 2002) or that
it represents response conflict (Yeung, Cohen, & Botvinick,
2004).
Although much less studied, the error positivity (Pe) is another ERP component observed on error trials. The Pe follows
the ERN as a positive deflection 200–400 ms after the commission of an error (Falkenstein, Hoormann, Christ, & Hohnsbein,
2000; Nieuwenhuis, Ridderinkhof, Blom, Band, & Kok, 2001;
Overbeek, Nieuwenhuis, & Ridderinkhof, 2005) and has a more
posterior midline scalp distribution than the ERN (Falkenstein
et al., 2000). It has been suggested that the Pe reflects error
awareness (Leuthold & Sommer, 1999; Nieuwenhuis et al.,
2001), the emotional assessment of the error (Falkenstein et al.,
2000), or a P300-like orienting response to errors (Davies,
Segalowitz, Dywan, & Pailing, 2001; Hajcak, McDonald, &
Simons, 2003).
Address reprint requests to: Greg Hajcak, Ph.D., Department of
Psychology, Stony Brook University, Stony Brook, NY 11794-2500,
USA. E-mail: [email protected]
957
958
signal-to-noise ratio (SNR) of the ERPs was also assessed. By
using a multitude of analytic methods, we sought to determine
how adding error trials would influence the quantification of
error-related brain activity.
Method
Participants
Seventy undergraduate students (43 female) participated in the
current study. Data from 2 participants were not used due to
excessive artifacts. Only participants who made at least 14 errors
(N 5 53; 33 female) were included. Additionally, for the Pe analysis, 1 participant did not have usable data from the electrode of
interest (Pz). No participants discontinued their participation in
the experiment once procedures had begun, and all participants
received course credit for their participation.
Task
The task was an arrow version of the flanker task (cf. Hajcak
et al., 2005). On each trial, five horizontally aligned arrowheads
were presented, and participants had to respond to the direction
of the central arrowhead by pressing the left or right mouse button. Half of all trials were compatible (‘‘ooooo’’
or ‘‘44444’’) and half were incompatible (‘‘oo4oo’’
or ‘‘44o44’’); all stimuli were presented for 200 ms with an
intertrial interval that varied randomly from 500 to 1000 ms.
Procedure
Following a brief description of the experiment, electroencephalographic (EEG) sensors were attached and the participant was
given detailed task instructions. Participants performed a practice block consisting of 30 trials and were told to try to be as
accurate and fast as possible. The actual experiment consisted of
eight blocks of 30 trials. To encourage both fast and accurate
responding, participants received feedback based on their performance at the end of each block. If performance was 75%
correct or lower, the message ‘‘Please try to be more accurate’’
was displayed; performance above 90% correct was followed by
‘‘Please try to respond faster’’; otherwise, the message ‘‘You’re
doing a great job’’ was displayed.
Psychophysiological Recording, Data Reduction, and Analysis
All EEG recording, filtering, and eye movement correction parameters were identical to those reported in Hajcak and Foti
(2008). Briefly, continuous EEG activity was sampled at 512 Hz
using an ActiveTwo head cap and the ActiveTwo BioSemi system
(BioSemi, Amsterdam, The Netherlands). Recordings were
taken from 64 scalp electrodes based on the 10–20 system, as
well as from 2 electrodes placed on the left and right mastoids.
Four electrodes recorded the electrooculogram generated from
eyeblinks and eye movements: Vertical eye movements and blinks
were measured with 2 electrodes placed approximately 1 cm
above and below the right eye; horizontal eye movements were
measured with 2 electrodes placed approximately 1 cm beyond
the outer edge of each eye. Off-line analysis was performed using
Brain Vision Analyzer software (BVA, Brain Products, Gilching,
Germany). EEG data were re-referenced to the numeric mean of
the mastoids and bandpass filtered with cutoffs of 0.1 and 30 Hz.
Eyeblink and ocular corrections were made using the method
developed by Gratton, Coles, and Donchin (1983). Specific intervals for individual channels were rejected in each trial using a
D.M. Olvet and G. Hajcak
semiautomated procedure, with physiological artifacts identified
by the following criteria: a voltage step of more than 50.0 mV
between sample points, a voltage difference of more than 300.0
mV within a trial, and a maximum voltage difference of less than
0.50 mV within a 100-ms interval. All trials were also visually
inspected for other artifacts.
Response-locked ERPs were computed for error trials. The
ERN was evaluated as the average activity in a 0–100-ms window
relative to response onset at FCz; the Pe was evaluated as the
average activity from 200 to 400 ms following response onset at
Pz. A 200-ms window prior to the response ( 400 to 200 ms)
served as the baseline. For all analyses, the ERN and Pe were
quantified based on a random subset of 2, 4, 6, 8, 10, 12, and 14
errors from each participant; the grand average ERN and Pe
were quantified based on all error trials. Area measures of the
ERN and Pe were chosen for two reasons: First, peak measures
might be especially sensitive to noise (see Luck, 2005); second,
within-subject measures such as SNR and Cronbach’s alpha necessitated the use of an area measure, and it was preferable to use
the same ERN/Pe metric across all analyses.
In all cases, the ERN and Pe were statistically evaluated using
SPSS (Version 16.0) General Linear Model software; Greenhouse–Geisser correction was applied to p values associated with
multiple degrees of freedom, repeated measures comparisons,
when appropriate. Paired-sample t tests were used to compare
the ERN and the Pe for each randomly chosen error trial averages (i.e., 2, 4, 6, 8, 10, 12, and 14) as 2 trials were increasingly
added to the average. Pearson’s correlation coefficient was also
used to examine the relationship between smaller error trial averages and the grand average ERN and Pe. Internal reliability of
the ERN and Pe as a function of increasing trial numbers was
quantified with Cronbach’s alpha. SNR at FCz and Pz was estimated using a process available in BVA. First, noise is estimated by summing the squares of the difference between each
data point and the average EEG value; this sum is then divided
by the number of data points minus one. Second, average total
power is estimated by taking the average of the squared values of
each data point. Average power of the signal then equals the
average total power minus the average noise power. SNR is then
calculated as the average signal power divided by average noise
power. SNR was assessed using a repeated measures ANOVA
using number of trials (2, 4, 6, 8, 10, 12, 14, and grand average)
as the within-subjects factor. Post hoc analyses were performed using paired-sample t tests, and significance levels were
adjusted with Bonferroni’s correction for multiple comparisons
(po.007).
Results
Between-Subject Comparisons
On average, participants made 27.47 (SD 5 8.15) errors while
performing the task. Figure 1 presents the average and standard
deviation of the ERN and Pe for random trial averages and for
the grand averages. ERPs for the ERN and the Pe from a representative subject for random trial averages and the grand average are presented in Figure 2. Paired-sample t-tests were
performed on both the ERN and the Pe area measures to examine differences between smaller trial averages as 2 trials were
increasingly added to the averages. There were no significant
differences when comparing increasing number of trials (2 vs. 4
trials, 4 vs. 6 trials, 6 vs. 8 trials, 8 vs. 10 trials, 10 vs. 12 trials,
ERN stability
Figure 1. The average (top) and standard deviation (bottom) of the ERN
at FCz (left ordinate) and the Pe at Pz (right ordinate) as a result of
adding two random trials to the average and for the grand average. Please
note that the left and right ordinates have different scales.
12 vs. 14 trials, and 14 vs. grand average; all ps4.05). The Pe for
all trials also did not differ (ps4.05). Figure 1 suggests that both
the average of the ERN and the Pe appear to stabilize after six
errors are included in the average. The standard deviation for
both components appears to decrease as more trials are added,
although the standard deviation appears fairly similar after
approximately 10 trials.
Additionally, we explored the relationship between each trial
average and the ERN/Pe grand average using the Pearson correlation coefficient. Figure 3 presents the correlation coefficient
between the grand average ERN/Pe and the ERN/Pe based on
fewer trials. All pairs were highly significant (po.001). Moreover, the ERN and Pe averages based on just six trials were
highly correlated with the grand average ERN and Pe (rs4.80).
959
Within-Subject Comparisons
Hinton, Brownlow, McMurray, and Cozens (2004) have suggested that Cronbach’s alpha exceeding .90 indicates excellent
internal reliability, between .70 and .90 indicates high internal
reliability, from .50 to .70 indicates moderate internal reliability,
and below .50 is low. Figure 4 presents Cronbach’s alpha for the
ERN as progressively more trials were considered. Consistent
with the impression from Figure 4, moderate and high internal
reliabilities for the ERN were obtained with 6 and 10 errors,
respectively. Cronbach’s alpha for the Pe as a function of trial
number is also presented in Figure 4. For the Pe, moderate and
high internal reliabilities were obtained with just 2 and 6 trials,
respectively. Thus, the ERN and Pe both demonstrated moderate internal reliability with just 6 errors and high internal reliability with 10 errors.
Estimates of the SNR for the ERN and Pe were also examined. SNR scores for the ERN starting with at least six error trials
ranged from .25 to .63. For the ERN, a repeated measures
ANOVA confirmed that there was a significant difference in the
SNR across averages, F(7,364) 5 15.06, po.001. Post hoc
paired-sample t tests indicated that there was a significant difference when comparing the ERN for 2 versus 4 trials, t(52) 5 3.61,
po.001, 6 versus 8 trials, t(52) 5 2.86, po.001, and 14 versus
grand average, t(52) 5 3.17, po.003; however, there was no
difference between the ERN for 4 versus 6 trials, t(52) 5 0.04,
p4.05, 8 versus 10 trials, t(52) 5 1.56, p4.05, 10 versus 12 trials,
t(52) 5 0.32, p4.05, and 12 versus 14 trials, t(52) 5 0.20,
p4.05. Thus, the SNR for the ERN did not change substantially
after 8 trials were added to the average.
SNR scores for the Pe starting with at least 6 error
trials ranged from 1.74 to 2.35. For the Pe, the SNR also
differed as more trials were added to the averages,
F(7,357) 5 19.25, po.001. Post hoc paired t tests showed that
there was a significant difference when comparing the Pe for 2
versus 4 trials, t(51) 5 3.58, po.001; however, there was no significant difference for 4 versus 6 trials, t(51) 5 1.29, p4.05, 6
versus 8 trials, t(51) 5 0.02, p4.05, 8 versus 10 trials,
t(51) 5 2.47, p4.01, 10 versus 12 trials, t(51) 5 0.28, p4.05,
12 versus 14 trials, t(51) 5 0.57, p4.05, and 14 versus grand
average, t(51) 5 2.34, p4.02. Therefore, the SNR for the
Pe did not significantly change once 4 trials were included in
the average.
Figure 2. ERPs from a representative single subject for trials containing the average of 2, 6, 10, and 14 randomly selected trials and the grand average
(i.e., 21 errors). ERP averages at FCz (top) and Pz (bottom) following error responses. Please note that the ERP figures have different scales.
960
D.M. Olvet and G. Hajcak
Figure 3. The Pearson correlation coefficient between the ERN at FCz
and Pe at Pz derived from averages based on increasing numbers of trials
with the grand average ERN and the grand average Pe, respectively.
Discussion
The present study examined ERPs among 53 individuals who
made approximately 27 errors on averageFeach made at least
14 errorsFto examine how measures of the ERN and Pe performed when calculated based on an increasing number of randomly chosen error trials. Analyses suggested that the ERN and
Pe became fairly stable once between six and eight trials were
included, depending on the metric examined. For instance, the
average of the ERN and Pe appeared to stabilize after about six
trials per participant were included in averages. Moderate internal reliability was also achieved with 6 errors. Moreover, in ERP
averages based on six trials per participant, the ERN and Pe were
highly correlated (r4.80) with the grand average ERN and Pe,
respectively. In addition, the SNR for a smaller number of trials
did not significantly differ much from the SNR for the grand
average after eight and four trials were added to the average for
the ERN and the Pe, respectively. Although it is possible that the
lack of significant differences is due to the small number of trials
in each average, we used multiple measures that all converged on
a common result.
In the present study, the ERN was elicited in a relatively brief
speeded reaction time task comprised of only 240 trialsFa task
that lasted less than 10 min. Out of a total of 70 participants, 64
made at least eight errors and 66 participants made at least six
errors. Thus, if one uses six to eight error trials as a minimum, less
than 10% of the original sample would need to be excluded.
Collectively, these data suggest that brief tasks can be used to
elicit a sufficient number of errors for quantifying the ERN and
Pe. This is particularly relevant in light of a growing literature
Figure 4. Cronbach’s alpha for the ERN at FCz (left ordinate) and the Pe
at Pz (right ordinate) as increasing number of trials were examined. Please
note that the left and right ordinates have different scales.
relating the ERN to certain psychiatric disorders. For instance,
evidence suggests that internalizing psychopathologies such as
anxiety (Gehring, Himle, & Nisenson, 2000; Hajcak et al., 2008;
Johannes et al., 2001; Ladouceur, Dahl, Birmaher, Axelson, &
Ryan, 2006) and depression (Chiu & Deldin, 2007; Holmes &
Pizzagalli, 2008) are characterized by increased ERNs, whereas
externalizing psychopathology is related to reduced ERNs (cf.
Olvet & Hajcak, 2008, for a review).
The current data suggest that six to eight trials are
adequate to reliably assess error-related brain activity, which
might aid in brief ERN assessments in clinical contexts. Additionally, it seems reasonable to include participants who make a
relatively small number of mistakes. These data further suggest
that it is possible to examine more infrequent error-related
phenomena (i.e., double errors; cf. Hajcak & Simons, 2008).
In light of the suggestion that an increased ERN may reflect a
stable trait-like marker for internalizing psychopathology (Olvet
& Hajcak, 2008), it will be important for future research to
also examine the test–retest reliability of error-related brain
activity.
In the present study, we used a number of statistical approaches to determine the stability of the ERN and the Pe. Future studies might further examine this issue using more
sophisticated statistical approaches (i.e., hierarchical linear modeling) and consider whether the current results generalize to other
tasks (e.g., Stroop) that are used in some studies to elicit the
ERN. Moreover, it will be important to independently examine
reliability measures in children to determine whether more
trials are required to obtain a stable ERN and Pe earlier in
development.
REFERENCES
Amodio, D. M., Harmon-Jones, E., Devine, P. G., Curtin, J. J., Hartley,
S. L., & Covert, A. E. (2004). Neural signals for the detection of
unintentional race bias. Psychological Science, 15, 88–93.
Band, G. P., & Kok, A. (2000). Age effects on response monitoring in a
mental-rotation task. Biological Psychology, 51, 201–221.
Chiu, P. H., & Deldin, P. J. (2007). Neural evidence for enhanced error
detection in major depressive disorder. American Journal of Psychiatry, 164, 608–616.
Cohen, J., & Polich, J. (1997). On the number of trials needed for P300.
International Journal of Psychophysiology, 25, 249–255.
Davies, P. L., Segalowitz, S. J., Dywan, J., & Pailing, P. E. (2001). Errornegativity and positivity as they relate to other ERP indices of
attentional control and stimulus processing. Biological Psychology,
56, 191–206.
Dehaene, S., Posner, M. I., & Tucker, D. M. (1994). Localization of a
neural system for error detection and compensation. Psychological
Science, 5, 303–305.
Falkenstein, M., Hohnsbein, J., Hoormann, J., & Blanke, L. (1991).
Effects of crossmodal divided attention on late ERP components. II.
Error processing in choice reaction tasks. Electroencephalography and
Clinical Neurophysiology, 78, 447–455.
Falkenstein, M., Hoormann, J., Christ, S., & Hohnsbein, J. (2000). ERP
components on reaction errors and their functional significance: A
tutorial. Biological Psychology, 51, 87–107.
961
ERN stability
Franken, I. H., van Strien, J. W., Franzek, E. J., & van de Wetering, B. J.
(2007). Error-processing deficits in patients with cocaine dependence.
Biological Psychology, 75, 45–51.
Gehring, W. J., Goss, B., Coles, M. G. H., Meyer, D. E., & Donchin, E.
(1993). A neural system for error detection and compensation. Psychological Science, 4, 385–390.
Gehring, W. J., Himle, J., & Nisenson, L. G. (2000). Action-monitoring
dysfunction in obsessive-compulsive disorder. Psychological Science,
11, 1–6.
Gratton, G., Coles, M. G., & Donchin, E. (1983). A new method for offline removal of ocular artifact. Electroencephalography and Clinical
Neurophysiology, 55, 468–484.
Hajcak, G., & Foti, D. (2008). Errors are aversive: Defensive motivation
and the error-related negativity. Psychological Science, 19, 103–108.
Hajcak, G., Franklin, M. E., Foa, E. B., & Simons, R. F. (2008). Increased error-related brain activity in pediatric obsessive-compulsive
disorder before and after treatment. American Journal of Psychiatry,
165, 116–123.
Hajcak, G., McDonald, N., & Simons, R. F. (2003). To err is autonomic:
Error-related brain potentials, ANS activity, and post-error compensatory behavior. Psychophysiology, 40, 895–903.
Hajcak, G., Moser, J. S., Yeung, N., & Simons, R. F. (2005). On the
ERN and the significance of errors. Psychophysiology, 42, 151–160.
Hajcak, G., & Simons, R. F. (2008). Oops! . . . I did it again: An ERP and
behavioral study of double-errors. Brain and Cognition, 68, 15–21.
Hinton, P. R., Brownlow, C., McMurray, I., & Cozens, B. (2004). SPSS
explained. East Sussex, UK: Routledge.
Holmes, A. J., & Pizzagalli, D. A. (2008). Spatiotemporal dynamics of
error processing dysfunctions in major depressive disorder. Archives
of General Psychiatry, 65, 179–188.
Holroyd, C. B., & Coles, M. G. (2002). The neural basis of human error
processing: Reinforcement learning, dopamine, and the error-related
negativity. Psychological Review, 109, 679–709.
Holroyd, C. B., Dien, J., & Coles, M. G. (1998). Error-related scalp
potentials elicited by hand and foot movements: Evidence for an
output-independent error-processing system in humans. Neuroscience
Letters, 242, 65–68.
Johannes, S., Wieringa, B. M., Nager, W., Rada, D., Dengler, R., Emrich, H. M., et al. (2001). Discrepant target detection and action
monitoring in obsessive-compulsive disorder. Psychiatry Research,
108, 101–110.
Ladouceur, C. D., Dahl, R. E., Birmaher, B., Axelson, D. A., & Ryan,
N. D. (2006). Increased error-related negativity (ERN) in childhood
anxiety disorders: ERP and source localization. Journal of Child Psychology and Psychiatry, 47, 1073–1082.
Leuthold, H., & Sommer, W. (1999). ERP correlates of error processing
in spatial S-R compatibility tasks. Clinical Neurophysiology, 110,
342–357.
Luck, S. J. (2005). An introduction to the event-related potential technique.
Cambridge, MA: MIT Press.
Luu, P., Collins, P., & Tucker, D. M. (2000). Mood, personality, and selfmonitoring: Negative affect and emotionality in relation to frontal
lobe mechanisms of error monitoring. Journal of Experimental Psychology, General, 129, 43–60.
Mathalon, D. H., Fedor, M., Faustman, W. O., Gray, M., Askari, N., &
Ford, J. M. (2002). Response-monitoring dysfunction in schizophrenia: An event-related brain potential study. Journal of Abnormal
Psychology, 111, 22–41.
Morris, S. E., Yee, C. M., & Nuechterlein, K. H. (2006). Electrophysiological analysis of error monitoring in schizophrenia. Journal of
Abnormal Psychology, 115, 239–250.
Nieuwenhuis, S., Ridderinkhof, K. R., Blom, J., Band, G. P., & Kok, A.
(2001). Error-related brain potentials are differentially related to
awareness of response errors: Evidence from an antisaccade task.
Psychophysiology, 38, 752–760.
Olvet, D. M., & Hajcak, G. (2008). The error-related negativity (ERN)
and psychopathology: Toward an endophenotype. Clinical Psychology Review, 28, 1343–1354.
Overbeek, T. J. M., Nieuwenhuis, S., & Ridderinkhof, K. R. (2005).
Dissociable components of error processing: On the functional significance of the Pe vis-a-vis the ERN/Ne. Journal of Psychophysiology, 19, 319–329.
Pailing, P. E., & Segalowitz, S. J. (2004). The effects of uncertainty
in error monitoring on associated ERPs. Brain and Cognition, 56,
215–233.
Polich, J. (1986). P300 development from auditory stimuli. Psychophysiology, 23, 590–597.
Ridderinkhof, K. R., Ullsperger, M., Crone, E. A., & Nieuwenhuis, S.
(2004). The role of the medial frontal cortex in cognitive control.
Science, 306, 443–447.
Ullsperger, M., & von Cramon, D. Y. (2006). How does error correction
differ from error signaling? An event-related potential study. Brain
Research, 1105, 102–109.
van Veen, V., & Carter, C. S. (2002). The timing of action-monitoring
processes in the anterior cingulate cortex. Journal of Cognitive
Neuroscience, 14, 593–602.
Yeung, N., Cohen, J. D., & Botvinick, M. M. (2004). The neural basis of
error detection: Conflict monitoring and the error-related negativity.
Psychological Review, 111, 931–959.
Zirnheld, P. J., Carroll, C. A., Kieffaber, P. D., O’Donnell, B. F., Shekhar, A., & Hetrick, W. P. (2004). Haloperidol impairs learning and
error-related negativity in humans. Journal of Cognitive Neuroscience,
16, 1098–1112.
(Received June 21, 2008; Accepted December 9, 2008)